Exploiting Majority Acceptable Arguments in Composite Ontology Matching

Abstract

Ontology matching consists of generating an alignment (set of correspondences) from a pair of ontologies. This process has been seen as a mainstream solution to the semantic heterogeneity problem in ontology-based systems. A wide diversity of matching solutions has been proposed, which exploit different features within an ontology. Matching systems usually differ in their results and an important issue is to combine different matching results and deal with potential conflicts that arise from the different views. Our approach exploits argumentation theory as a way for dealing with that issue. Here, arguments are as positions that support or reject correspondences and argumentation frameworks support the creation and exchange of arguments, followed by the reasoning on their acceptability. First, match- ers generate their correspondences and represent them as arguments. Next, they share their arguments and interpret them on the basis of argumentation frameworks and individ- ual preferences. As a result, each matcher has a subset of acceptable arguments, from the set of arguments initially shared. The subset of globally acceptable arguments (consensus) is computed from the individual. In this paper, we exploit the notion of majority, where argu- ments being acceptable by the majority of matchers are considered as a consensus on the initial alignments. We evaluate our proposal on a standard set of alignments. Considering the correspondences represented as arguments acceptable for the majority of individual subsets, both precision and recall are improved, specially when compared with the subsets acceptable for every matcher or for some matchers.